From: Stephen van Vuuren <stephen@sv2studios.com>
To: After Effects Mail List <AE-List@media-motion.tv>
Subject: RE: [AE] Using optical flow to speed up 3D renders
Thread-Topic: [AE] Using optical flow to speed up 3D renders
Thread-Index: AQHPOGjU754rNPI4aEKrRxJybMwO+JrSvzwA
Date: Wed, 5 Mar 2014 17:34:32 +0000
Message-ID: <d1b52ed1b41e43a4901e0a08d6b8be2e@CO1PR06MB265.namprd06.prod.outlook.com>
References: <list-5398193@media-motion.tv>
In-Reply-To: <list-5398193@media-motion.tv>
Accept-Language: en-US
Content-Language: en-US
received-spf: None (: sv2studios.com does not designate permitted sender
 hosts)
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

I have not used optical flow for speeding up 3D renders specifically but I'=
ve used Twixtor for a number of projects over the years with various types =
of sources that have included some 3D source material.

The problem with optical flow is that it's not an automatic solution i.e. s=
et and forget as camera motion, foreground and background layers can confus=
e and cause subtle or obvious artifacts.=20

On say a two second shot with a lot of camera and subject movement (especia=
lly if CGI has handheld or zoom camera simulation), it's likely manual twea=
king of the optical flow is going to be necessary even with tweaked motion =
vectors.

If the 3D CGI renders are fairly static and predictable (for the optical fl=
ow) it's possible to get settings that will work for a whole shot.=20

But the reason optical flow is simply not a default solution is that ultima=
tely computers and software is still stupid and can't compensate for unexpe=
cted situations.

And while it's much faster than it used to be, unless the CGI renders are b=
rutal, the renders are still going to be considerable with optical flow req=
uired to get artifact free processing. Plus manual labor time.

Again - certain controlled scenarios it might work - but unlikely to be a s=
imple plug and play solution.

stephen van vuuren
336.202.4777

http://www.insaturnsrings.com/
http://www.sv2dcp.com/
http://www.sv2studios.com/

A film is - or should be - more like music than like fiction. It should be =
a progression of moods and feelings. The theme, what's behind the emotion, =
the meaning, all that comes later.
-Stanley Kubrick

-----Original Message-----
From: After Effects Mail List [mailto:AE-List@media-motion.tv] On Behalf Of=
 Chris Zwar
Sent: Wednesday, March 05, 2014 6:48 AM
To: After Effects Mail List
Subject: [AE] Using optical flow to speed up 3D renders

I have this vague recollection that optical flow technology (e.g. twixtor a=
nd reel smart motion blur) was originally developed to help speed up 3D ren=
dering workflows.  I think the idea was that 3D animators could render out =
every 2nd or 3rd frame, and use an optical flow plugin like twixtor to crea=
te the in-betweens.  Even if that's not strictly correct, the potential is =
there.  Some of the photo-realistic projects I've worked on have had 3D ren=
der times of 3 - 5 hours a frame.  Rendering every 2nd frame is effectively=
 halving the overall render time, which can be a massive saving.  Even a sl=
ow After Effects plugin is usually only seconds per frame, not hours.

So I was wondering if anyone has actually done this, or tried using other 2=
D techniques to help speed up 3D rendering. =20

I can think of 3 ways in which slow 3D renders can be compensated for by fa=
ster compositing techniques:

1) Up-resing.  For example rendering at 720p instead of 1080p and scaling u=
p the finished renders.  If compositing multiple passes, only the slow rend=
ers need to be smaller and scaled up.
2) De-noising.   Forgive me for not knowing the correct terminology, but wh=
en rendering with global illumination it seems that there's an overall qual=
ity setting that directly determines both the speed of rendering and the no=
isiness of the image.  Rendering with a lower setting can make renders nois=
ier, but a de-noising plugin such as Neat Video can fix this.
3) As stated above, rendering every 2nd or 3rd frame and using something li=
ke twixtor to create the missing frames.  A motion vector pass would make t=
his more accurate.

So I'm familiar with 2 of those 3 approaches - I have worked in situations =
where 3D passes are rendered at smaller sizes and then scaled up.  It works=
 very well and the time savings can be dramatic when dealing with very long=
 renders.  3D renders can be so clean that they scale up very well.

I have also worked in situations where the neat video de-noiser was used to=
 compensate for noisy GI renders, and again the savings can be dramatic - i=
n some cases this can almost half 3D rendering times.  Neat video seems to =
be an incredible plugin, so much faster and so much better than the AE equi=
valent.

So that leaves the optical flow technique as the one I haven't tried yet.  =
Has anyone done this?  I'd love to hear from real-world examples where peop=
le were able to render every 2nd or 3rd frame.  Is a motion vector pass ess=
ential for it to work properly?

Any other thoughts or insight welcome...

-Chris
+---End of message---+
To unsubscribe send any message to <ae-list-off@media-motion.tv>